Rough sets for finite mixture model based HEp-2 cell segmentation

Abstract

Automatic extraction of HEp-2 cells from an image is one key component for the diagnosis of connective tissue diseases. The gradual transition between cell and surrounding tissue renders this process difficult for any computer aided diagnostic systems. In this regard, the paper presents a new approach for automatic HEp-2 cell segmentation by incorporating a new probability distribution, called stomped normal (SN) distribution. The proposed method integrates judiciously the concept of rough sets and the merit of the SN distribution into an finite mixture model framework to provide an accurate delineation of HEp-2 cells. The intensity distribution of a class is represented by SN distribution, where each class consists of a crisp lower approximation and a probabilistic boundary region. Finally, experiments are performed on a set of HEp-2 cell images to demonstrate the performance of the proposed algorithm, along with a comparison with related methods.